Reducing work disability among the working population and particularly among older workers is an important issue on the scientific and policy agenda in many countries. The fraction of workers on disability insurance (DI) is vastly different across countries with similar economic development and comparable access to medical technology and treatment. Recent survey data show that significant differences between countries are also found in self-reports of work limiting disabilities. In comparing such self-reports, account should be taken of measurement issues such as differences in question wordings, justification bias and other differences in response scales that may exist between and within countries. This project aims at explaining differences in self-reported work limiting disability prevalence between several countries and between demographic groups within countries. Moreover, it aims at analyzing the dynamics of work disability and labor market status. We focus on measurement issues and several types of reporting bias in work limiting disability. We will investigate the role of health conditions, financial incentives and generosity of the DI program, and the importance of reference groups and social norms. The results will be used to construct static and dynamic models that identify reporting biases in work limiting disability and relate work-limiting disability to an individual's actual labor market status. The availability of innovative data in Internet surveys and the possibility of conducting a series of experiments make it possible to achieve these goals. Our main tool will be the use of vignettes that describe hypothetical persons with various kinds of work limiting disabilities. Respondents are asked to evaluate the work disability of the vignettes on the same scale as their won work disability. This will help us to identify reporting bias in cross-section and panel data models of work limiting disability and labor market status. Cros-section and panel data models will be developed that incorporate framing and justification bias and allow for different reporting biases in several domains of work limiting health, such as affect, pain and heart problems. Experiments with different formulations of the vignette will be used to test and relax assumptions on how to interpret the vignette evaluations. Moreover, the 1995 reforms of the UK disability system will be used to identify and analyze reporting bias in an alternative way.